LC-MS/MS method for simultaneous quantification of dexamethasone and tobramycin in rabbit ocular biofluids

Abstract

The complexity of Tobradex® ointment formulation (dexamethasone 0.1 wt% and tobramycin 0.3 wt%) and the high cost of pharmacokinetic (PK) studies in human aqueous humor prevent generic drug companies from moving forward with a Tobradex®-equivalent product development. The in vitro drug release test would be an alternative approach for differentiating the generic formulations containing both DEX and TOB, and the results should be correlated with the in vivo ocular PK studies for further evaluation. To facilitate the in vivo ocular PK studies, a sensitive, rapid and specific liquid chromatography-tandem mass spectrometry (LC-MS/MS) method that can simultaneously quantify both dexamethasone (DEX) and tobramycin (TOB) in rabbit ocular matrices including tear, aqueous humor and cornea was established and validated. The lower limit of quantification (LLOQ) was 1.5 ng/ml for DEX and 3 ng/ml for TOB with good precision and accuracy. Both intra- and inter-batch precisions were within ±15%, and the accuracy for all QCs was within the range of 85% – 115%. This new method was successfully applied for a pilot pharmacokinetic analysis of DEX and TOB in rabbit tears after topical administration of Tobradex® ointment.

1. Introduction

Tobradex® ophthalmic ointment (Alcon Laboratories Inc, Fort Worth, Texas) is a combination of dexamethasone (0.1wt%) and tobramycin (0.3wt%) for treating steroid-responsive inflammatory ocular conditions for which a corticosteroid is indicated and where superficial bacterial ocular infection exists [1]. There are no generic products available on the market, even though Tobradex® has been off patent since 2009. The complexity of Tobradex® ointment formulation and the high cost of pharmacokinetic (PK) studies in human aqueous humor present inherent challenges for generic drug companies developing Tobradex®-equivalent products. Additionally, the PK study in aqueous humor cannot accurately reflect topical drug release profiles due to the limited corneal drug permeability, especially for the hydrophilic Tobramycin. Considering the complications of in vivo PK studies in human, the in vitro assessment of generic formulations, and the correlation studies between the in vitro drug release test and the in vivo ocular PK (IVIVC) in rabbits can help generic drug development.

Topical ophthalmic formulations can directly deliver the drug to the local ocular tissues. Due to the complex anatomical structure of the eye and the existence of various ocular barriers, most of drugs following topical administration are released into the tears and some of drugs reach into the aqueous humor, but the drugs reaching to the back of the eye are very limited [2]. Thus, the pharmacokinetic studies of DEX/TOB in rabbit tears, aqueous humor and cornea will be most important for the IVIVC studies. To this end, a rapid, sensitive and selective analytical method that can simultaneously quantify both dexamethasone (DEX) and tobramycin (TOB) in rabbit tear fluids, aqueous humor and cornea is required, especially since the in vivo animal PK analysis are challenged by the limited sample amount and the low drug content in rabbit cornea and aqueous humor. DEX is a hydrophobic corticosteroid with LogP 1.83 (Fig. 1A), and reverse phase liquid chromatography tandem mass spectrometry (LC-MS/MS) methods have been developed for DEX quantification in biological samples [3–5]. TOB is a hydrophilic, aminoglycoside antibiotic with LogP −5.8 (Fig. 1B) lacking of chromophores that make it challenging for quantitative analysis. Numerous assays have been reported using derivatization with UV, fluorescence or electrochemical detection methods [6–9]. However, derivatization methods are complicated and may not be compatible with DEX analysis. Moreover, the sensitivity of these methods is relatively low and cannot meet the sensitivity requirement for ocular tissue bioanalysis. LC-MS/MS methods have been applied for TOB analysis in biological samples with a limit of quantification more than 10 ng/ml [10, 11]. Considering the hydrophilicity of TOB, hydrophilic interaction chromatography (HILIC) method would be a suitable approach for TOB analysis. However, HILIC method is not applicable for DEX analysis considering the different characteristic of DEX and TOB. Due to the high polarity, aminoglycosides are not retained in conventional C18 columns that makes it even more difficult to separate DEX and TOB at the same time. We applied a pentafluorophenyl (PFP) reverse phase column to increase the selectivity for TOB and DEX. Additional TFA in the aqueous phase is applied to facilitate the retention of TOB on the column. Here, we established and validated a rapid, selective and sensitive method that can simultaneously quantify DEX and TOB in rabbit tears, cornea and aqueous humor. The developed method has been validated according to the FDA guidance on Bioanalytical Method Validation [12]. The developed method has been successfully applied for the pilot pharmacokinetic study of DEX and TOB in rabbit tears after topical administration of branded Tobradex® ointment. It would be further applied to pharmacokinetic studies of DEX and TOB in rabbit cornea and aqueous humor in the future.

Fig. 1:

Chemical structure of (A) dexamethasone (DEX) and its internal standard (C) dexamethasone-d5 (DEX-d5); (B) tobramycin (TOB) and its internal standard (D) sisomicin (Siso).

2. Experimental

2.1. Chemicals and materials

Dexamethasone (DEX, Lot: D298800, purity 98%), Dexamethasone-d5 (DEX-d5, Lot: D298802, purity 98%), tobramycin (TOB, Lot: T524000, purity 98%) and sisomicin (siso, Lot: S488505, purity 96%) were purchased from Toronto Research Chemicals (North York, ON, CA). Methanol, acetonitrile and water at LC-MS grade were obtained from Fisher Scientific (Waltham, MA, USA). Formic acid, TFA, ammonium acetate, sodium bicarbonate, calcium chloride and sodium chloride were purchased from Millipore-Sigma (Burlington, MA, USA).

2.2. Chromatographic and mass spectrometric conditions

LC-MS/MS analysis was performed on an Acquity UPLC I Class system coupled to a Xevo TQS micro Triple Quadrupole Mass Spectrometry system equipped with an electrospray ionization (ESI) interface from Waters (MA, USA). The LC system includes an autosampler, a binary pump, analytical column and two 6-port switching valves. The chromatographic separation was achieved on Phenomenex (Torrance, CA) Kinetex PFP column (2.1 × 100 mm, 1.7 μm), with column temperature set at 40°C. Mobile phase A was 10 mM ammonium acetate buffer solution containing 0.14% (v/v) TFA. Mobile phase B was acetonitrile containing 0.5% (v/v) formic acid. Sample separation was conducted under gradient conditions with a flow rate of 0.3 ml/min. Initial elution composition was 95% (v/v) mobile phase A for 0.5 min, followed by a linear gradient up to 20% (v/v) mobile phase A at 3.5 min, held for 0.5 min, returning back to initial composition at 4.1 min. Column equilibration time was 1.9 min, leading to a total run time of 6 min.

The online diversion method was developed to limit the amount of TFA entering into the MS source and to provide the additional online sample clean up. The online diversion was achieved by changing the configuration of the two six-port valves at different time points to control the flow direction of the mobile phases (Fig. 2). From 0 – 2 min, the valve was in position A, 20 μl sample was loaded to the PFP column by a loading pump (pump 1). The target compounds were retained on the column and water-soluble impurities in the matrices including salts were directed to the waste by the high ratio of aqueous mobile phase. From 2.0 – 4.1 min, the valve position was switched to position B where the PFP column was connected to the MS source. The analytes that retained on the column were flushed out from the column by pump 2 and entered the MS source for analysis. Subsequently, the valve was switched back to position A. Valve switching events as well as the mobile phase gradients are shown in Table 1. The retention time of DEX, TOB, DEX-d5 and sisomicin was 3.08, 2.88, 3.07 and 2.75 min respectively.

Fig. 2:

Schematic workflow of the online diversion LC-MS/MS system. (A) the mobile phases were diverted to the waste. (B) the PFP column was connected to the MS source.

Table 1.

Mobile phase gradient and valve switching. A: aqueous phase. B: organic phase. Flow rate is 0.3 ml/min.

Time/min	A% (v/v)	B% (v/v)	Valve position
0	95	5	A
0.5	95	5	A
3	20	80	0.5–2 min: A; 2–3 min: B
4	20	80	B
4.1	95	5	A
6	95	5	A

The ESI was operated in positive mode and multiple reaction monitoring (MRM) was applied for detection of analytes. The optimized MS parameters were: source temperature, 150 °C; capillary voltage, 3kV; desolvation temperature 350 °C; desolvation gas flow, 600 L/Hr; cone gas flow, 100 L/Hr; dwell time, 0.046s. The pure nitrogen was applied for cone gas and desolvation gas, and argon was used for collision gas flow. The MRM transitions and compound dependent parameters cone voltage and collision energy were optimized at 32 V and 15 V for DEX (393.08>355.13), and 30V and 20 V for TOB (468.20>163.10). MRM transitions and compound dependent parameters for internal standards, DEX-d5 (398.15>360.14) were 34 and 12V, and sisomicin (448.32>160.00) were 35 and 20V. The data acquisition was performed by using Mass Lynx software and processing was performed by Target Lynx program.

2.3. Preparation of stock solutions, calibration curve standards and quality control samples.

Stock solutions of DEX and DEX-d5 (0.5 mg/ml) were prepared in methanol. Stock solutions of TOB and sisomicin (0.5 mg/ml) were prepared in water. All the stock solutions were stored at −20 °C before use. Working solutions containing both TOB and DEX (TOB: DEX=2:1) were prepared by diluting DEX and TOB stock solutions with methanol-water (70:30, v/v), and were used for preparation of calibration standards and quality control (QC) samples. Internal standard (IS) working solutions containing both sisomicin and DEX-d5 (sisomicin: DEX-d5 =2:1) were prepared by diluting sisomicin and DEX-d5 stock solutions with 70% methanol-water.

The artificial tear solution is comprised of 2 mg/ml sodium bicarbonate, 0.8 mg/ml calcium chloride and 6.7 mg/ml sodium chloride. Calibration standards containing 3, 4, 10, 20, 50, 100, 200, 300 ng/ml TOB and 1.5, 2, 5, 10, 25, 50, 100, 150 ng/ml DEX were prepared by spiking TOB/DEX working solutions with blank artificial tear solution. QCs were prepared in the similar way by diluting TOB/DEX working solutions with the blank matrix. QCs in artificial tear solution were prepared at concentrations of 3, 9, 125 and 250 ng/ml TOB containing 1.5, 4.5, 62.5 and 125 ng/ml DEX respectively. QCs in aqueous humor and cornea were prepared at concentrations of 25 and 125 ng/ml TOB containing 12.5 and 62.5 ng/ml DEX respectively.

2.4. Sample preparation

A simple protein precipitation method was used for DEX and TOB extraction from artificial tear solution, aqueous humor, and cornea. Blank aqueous humor and cornea tissues were collected from the dissected healthy adult NZW rabbit eyeballs (purchased from Pel-Freez Biologicals, LLC). Corneal tissues were cut into small pieces, wrapped in an aluminum foil, quickly frozen in liquid nitrogen and then immediately smashed with a hammer. The smashed cornea tissues were further homogenized in water for 5 min using Bullet blender® machine (Laboratory Supply Network, NH, USA). A 50 μl sample aliquot (artificial tear solution, aqueous humor and homogenized rabbit cornea matrices) was placed in 1.5 ml microcentrifuge tube, followed by adding of 20 μl IS (2.5 μg/ml sisomicin and 1.25 μg/ml DEX-d5) and vortex-mixing for 30s. The mixture was extracted with 150 μl cold acetonitrile-water (70:30, v/v) solution by vortex-mixing for 10 min. The supernatant was collected after centrifugation at 15,000× g for 10 min. 20 μl supernatant was injected into the LC system for analysis.

2.5. Method validation

The method for artificial tear solution was fully validated for selectivity, linearity, LLOQ, precision, accuracy, recovery, matric effects, stability and dilution integrity according to guidelines set by the United States Food and Drug Administration (FDA) for bioanalytical method validation [13]. Due to the limited amount of blank aqueous humor and cornea samples, the method for aqueous humor and cornea is partially validated, including the precision, accuracy, recovery, matrix effects and stability. The QCs of aqueous humor and cornea were quantified under the calibration curve of artificial tear solution.

The method selectivity was assessed by analyzing six lots of blank matrix including artificial tear solution, rabbit aqueous humor and cornea. Each individual lot was processed and analyzed as described above. The response of the background noise at the retention time of DEX/TOB was considered acceptable if it was less than 20% of the response of the LLOQ.

Each calibration curve (CC) consisted of one blank matrix sample, one blank matrix spiked with both ISs, and eight calibration samples covering the range of 3 to 300 ng/ml for TOB, and 1.5 to 150 ng/ml for DEX. To validate the linearity of the method, calibration curves were prepared on three different days with two replicates of each standard point. The calibration curves for both DEX and TOB were obtained by plotting the peak-area ratios of analytes to the IS (y) versus the concentration of analytes (x), with a weighing factor of 1/x². The slope, intercept and the correlation coefficient were determined and the residuals of each calibration standards were evaluated. The CCs were considered acceptable if the residuals were found within ±15% limit of theoretical concentrations (±20% for LLOQ). The LLOQ of the method was defined as the lowest concentration with the signal to noise ratio (S/N) of more than 10 and can be quantified with accuracy within ± 20% deviation from nominal value (DFN (%)) and relative standard deviation (RSD) of ± 20%.

Precision, accuracy, intra- and inter- batch reproducibility were evaluated using QC samples covering the calibration range. Intra-batch precision and accuracy were evaluated from QC samples at different concentrations on the same day. Inter-batch precision and accuracy evaluation were conducted through analyzing QC samples at three consecutive days. The RSD of QCs was calculated to estimate the precision. Accuracy was determined by comparing the calculated mean concentrations with the known concentrations. The accuracy for all QCs should be within 85%−115% of nominal concentration (80%−120% for LLOQ), and the precision should be within ± 15% (20% for LLOQ).

The extraction recovery of DEX and TOB from biological matrices were evaluated by comparing the mean area of the extracted QCs with that of samples at the same concentration obtained by spiking the standard solution with the extracted blank matrix samples. The recovery of internal standard was evaluated at the working concentration (DEX-d5: 125 ng/ml, sisomicin: 250 ng/ml) in the similar way. To evaluate the matrix effect (ME), blank matrices were extracted and spiked with the QC samples (post-spike sample). The ME was calculated by comparing the peak area of DEX/TOB in post-spike samples with that in external (unextracted) samples at the same concentration, the detailed equation is:

$$\text{ME}(\%)=(\frac{\text{peak}\text{area}\text{of}\text{analytes}\text{in}\text{post}-\text{spike}\text{sample}}{\text{peak}\text{area}\text{of}\text{analytes}\text{in}\text{unextracted}\text{sample}}-1)\times 100$$

Stability experiments were conducted to evaluate the stability of the TOB/DEX under different storage conditions. Freeze-thaw stability and post-preparative stability were evaluated. The freeze-thaw QCs were injected after three freeze-thaw cycles at −20°C with 24h interval between two cycles. The post-preparative stability was evaluated after storage of samples in the auto-sampler for 24 h at 4°C. The stability QCs were analyzed against freshly prepared calibration standards. Samples were considered stable if the concentration of DEX/TOB was within ± 15% of DFN (%) for the tested QC samples.

Due to the high concentration of DEX/TOB in tear samples, a dilution integrity test was performed to test the feasibility of the method for quantifying samples with TOB and DEX concentration higher than the upper concentration limit of the calibration range. For this purpose, the artificial tear solution containing 125 μg/ml TOB and 62.5 μg/ml DEX was prepared. The solution was diluted by 500-fold. The concentrations of DEX and TOB were calculated through the calibration curve.

2.6. Pilot animal study

All experimental protocols were approved by the VCU Animal Care and Use Committee. Three New Zealand White rabbits (2–3 kg) with mixed gender were purchased from Robinson Servic, Inc (Mocksville NC) for a pilot PK study to measure drug levels in the tear fluids (n=6 eyes). Rabbits were individually housed with free access to water and food. Approximately 50 mg of Tobradex® ointment (Alcon, Lake Forest, CA) was placed inside the center of the lower eye lid of rabbit using a spatula under minimal physical restrains. The lower eyelid was gently moved upwards to spread the dose uniformly over the corneal surface. Rabbits were held for 2 more minutes after the topical application to prevent them from shaking their head or pawing away the doses. The whole procedure was last less than 5 min with rabbits under minimal physcial restraint. Tears were collected through tear strips at 15 min, 30 min, 1 h, 2 h, 4 h, 6 h and 8 h post-dose under the minimal physical restraint. The amount of tears collected were calculated from the difference in tear strip weight before and after sample collection. The tear samples (n=6) were then weighed, and processed as described above.

2.7. Pharmacokinetic calculations

The pharmacokinetic (PK) parameters including Cmax, Tmax and AUC_(0−∞) were calculated by non-compartment analysis, using Phoenix version 7 (Pharsight Corporation, California, USA). The linear trapezoidal with a linear interpolation method was applied for calculation of AUC_(0−∞).

3. Results and discussion

3.1. Selection of IS

Since the deuterated form of the analytes are the most suitable IS for LC/MS/MS analysis, DEX-d5 (Fig. 1C) was selected as the IS for DEX. However, the deuterated form of TOB with high purity was not available. Therefore, the compound with similar characteristic of TOB was considered. Sisomicin (LogP −4.3; Fig. 1D) is an aminoglycoside with similar structure as TOB, and it has been applied as IS standard [14, 15]. Thus, sisomicin was selected as IS for TOB analysis in this method due to its similar physicochemical properties, ionization and retention time as TOB.

3.2. Mass spectrometry condition optimization

For the development of mass transitions (m/z), individual analytes and ISs prepared in methanol-water (50:50, v/v) solution at 500 ng/ml were infused by a syringe pump at a flow rate of 10 μl/min. The MS signal was optimized in full scan mode in the m/z range of 100–600 under positive ionization mode. The best signal of [M+H]⁺ ions for DEX at m/z 393.08 and [M+H]⁺ ion for TOB at m/z 468.20 were selected. After fragmentation, the analytical MRM at 390.85>355.13 for DEX and 468.20>163.12 for TOB were selected since they were the most abundant transitions, which were also reported in the previous studies [5, 14]. The analytical transitions at 398.14>360.13 for DEX-d5 and 448.32>160.00 for sisomicin were selected.

3.3. Chromatographic condition optimization

DEX is a hydrophobic drug with logP of 1.83, while TOB is a hydrophilic drug with logP of −5.8 (Fig. 1). The extremely different characteristics of the two drugs make it difficult to separate them within the same run. Meanwhile, the high sensitivity requirements and the limited amount of rabbit ocular tissues make it even more challenge for the LC method development. To develop a sensitive LC-MS/MS method for simultaneous quantification of DEX and TOB in rabbit ocular samples, various columns were attempted, including Acquity C18 column, BEC HILIC column and the Kinetex PFP column. C18 columns have been applied for DEX analysis [5], while TOB were not retained well on C18 column and multiple peaks were shown even with the help of TFA. HILIC column provided good retention and a single sharp peak for TOB separation, however the HILIC column cannot provide a suitable chromatography for DEX analysis. Compared with C18 and HILIC column, PFP column is the most suitable column for separation of DEX and TOB. Since mobile phase containing formic cannot provide a good separation for TOB, 0.14% TFA was added in the aqueous phase. TFA works as an ion-pairing reagent to increase the TOB retention on the PFP column, and the ion-pairing reagents (e.g. HFBA, TFA, PFPA) have been widely used for tobramycin and aminoglycoside analysis [15–18]. Considering the ion suppression effect caused by TFA, 0.5% formic acid was added to the organic phase to increase the ionization efficiency. Volatile salt was also employed to increase the ion strength. Finally, a gradient elution using a combination of 10 mM ammonium acetate buffer solution with 0.14% TFA and 0.5% formic acid acetonitrile solution was optimized to achieve the best sensitivity, efficiency, and peak shape for the separation of DEX and TOB. The flow rate was 0.3 ml/min, and the injection volume were 20 μl. Since TFA can cause ion suppression and source contamination [19], the online diversion method was developed to limit the amount of TFA entering the MS system. The mobile phase was diverted to the MS source from 2 to 4 min during the 6-min run. As a result, the total amount of TFA that enter the MS source was reduced by two-fold. Moreover, the online diversion method was coupled with the mobile phase gradient to conduct the additional online sample clean up. From 0–2 min, the analytical column was connected to the waste and the water-soluble salts were removed from the system under the high aqueous phase (Fig. 2).

Since the in vitro drug release medium has the similar composition of artificial tear solution, the analytical method can also be applied for the in vitro drug release sample analysis in the future. The representative chromatography of analytes and IS were shown in Fig. 3D and Fig. 4D. The retention time of DEX, DEX-d5, TOB and sisomicin was 3.08, 3.07, 2.75 and 2.88 min, respectively. It indicated that the online diversion did not influence the elution of analytes and IS.

Fig. 3:

Representative chromatograms of DEX and DEX-d5 in (A) blank artificial tear solution, (B) blank aqueous humor, (C) blank cornea matrix, and (D) blank artificial tear solution spiked at 1.5 ng/ml DEX (LLOQ) with 125 ng/ml DEX-d5.

Fig. 4:

Representative chromatograms of TOB and sisomicin (siso) in (A) blank artificial tear solution, (B) blank aqueous humor, (C) blank cornea matrix, and (D) artificial tear solution spiked at 3 ng/ml TOB (LLOQ) with 250 ng/ml sisomicin.

3.4. Optimization of extraction method

Due to the different characteristics of DEX and TOB, liquid-liquid extraction and reversed phase SPE cartridges may not be suitable for separation of both DEX and TOB. A simple protein precipitation method was explored for DEX and TOB extraction from biological matrices. Methanol and acetonitrile have been applied for DEX extraction from plasma [5]. Due to the high solubility of TOB in water (94 mg/ml), different ratios of water-methanol/acetonitrile solutions were evaluated for DEX and TOB extraction. Eventually, 70% acetonitrile in water was selected for DEX and TOB extraction from biological matrices since it provided the highest recovery for both analytes (more than 85% recovery for both DEX and TOB). The protein precipitation method was able to extract both DEX and TOB from the biological matrices at the same time with good recovery and did not require additional sample extraction procedures.

3.5. Method validation

The representative blank chromatograms of analytes and IS in artificial tear solution, aqueous humor and cornea matrix are shown in Fig. 3 A, B, C and Fig. 4 A, B, C. In all the three matrices, the responses of blank samples at the retention time of DEX/TOB were less than 20% of the responses in the LLOQ samples. The responses of blank samples at the retention time of DEX-d5/sisomicin were less than 5% of the responses at 125ng/ml DEX-d5 and 250 ng/ml sisomicin, respectively. Meanwhile, no interfering peaks at the retention time of analytes and IS were detected in the blank samples, which indicated that the method was able to selectively separate DEX and TOB without the interferences by other compounds. The calibration curve of DEX was linear as a function of concentration over the range of 1.5–150 ng/ml. The corresponding equation is y=0.0044*x+0.0015. The quadratic regression was applied for the regression of TOB covering the range of 3–300 ng/ml. The corresponding equation is y=−2.37e⁻⁷*x²+0.01*x-0.90e⁻³. Accuracy and precision at low concentration were significantly improved by applying 1/x² weighing factor. The representative chromatograms of DEX and TOB at LLOQ are shown in Fig. 3D and Fig. 4D. All the calibration standards presented residuals of ±15% from nominal concentration, confirming the suitability of the regression method. The LLOQ of TOB and DEX was 3 ng/ml and 1.5 ng/ml, respectively. The LLOQ of both drugs were reproducible with precision and accuracy within the FDA guidance recommended criteria with S/N more than 10 (Fig. 3D and Fig. 4D). The method enabled simultaneous quantification of TOB and DEX with an LLOQ for TOB that was three times lower than those were previously reported [6, 14, 20]. TFA in the aqueous phase compromised the DEX signal, however the sensitivity of DEX still was able to reach to 1.5 ng/ml which was enough for DEX quantification in rabbit aqueous humor after Tobradex® administration [5]. TFA in the mobile phase increased TOB retention on the column and didn’t cause significant influence on the sensitivity of DEX analysis.

The accuracy and precision of DEX and TOB in artificial tear solution are shown in Table 2. Six replicates of QCs in artificial tear solution at four different concentrations were prepared and analyzed for this study. Intra-batch precision and accuracy were evaluated in the same day, and inter-batch precision and accuracy were assessed from the analysis of the QC samples at 3 different days. The intra- and inter-batch accuracy were within 85%−115% of the nominal concentration, and the intra- and inter-day precision were within ±15%, which meets the acceptable limits of FDA guidance.

Table 2.

Intra- and inter- batch precision and accuracy of DEX and TOB in artificial tear solution with six replicates.

Analyte	Nominal Concentration ng/ml	Intra-batch		Inter-batch
		Accuracy (DFN, %)	Precision (RSD, %)	Accuracy (DFN, %)	Precision (RSD, %)
DEX	1.5	100.8	9.3	99.1	7.2
	4.5	105.3	4.5	98.6	8.3
	62.5	105.7	3.9	100.9	9.5
	125	103.4	7.9	98.2	8.0
TOB	3	98.8	12.3	110.0	14.8
	9	94.8	8.5	90.1	11.0
	125	102.7	3.1	102.0	5.6
	250	100.2	4.4	103.9	4.7

The extraction recovery and matrix effect of DEX and TOB from artificial tear solution, aqueous humor, cornea is summarized in Table 3. The simple protein precipitation method provided 85%−115% of DEX and TOB recovery from all the three matrices. The average recovery of DEX-d5 from artificial tear solution, cornea matrix and aqueous humor were 102.0%, 96.2% and 110%, respectively. The average recovery of sisomicin from artificial tear solution, cornea matrix and aqueous humor were 104.2%, 85.4% and 112.0% respectively. No significant matrix effects were shown (within ±15%). The internal standard response in calibration curve samples (CCs), QCs and tested rabbit tear samples were shown in Fig. S1 (supporting information). The response of DEX-d5 and sisomicin in CCs, QCs and tested rabbit tear samples were similar indicating the minimum matrix effects. Sisomicin and DEX-d5 responses in all the QCs and tested rabbit tear samples were within 85%~115% of the average internal standard responses in CCs. The protein precipitation method was proved to successfully extract both DEX and TOB from ocular tissues at the same time.

Table 3.

Recovery and matrix effects of DEX and TOB in artificial tear solution, rabbit aqueous humor and cornea.

	DEX			TOB
	Nominal Concentration ng/ml	Recovery (%)	ME (%)	Nominal Concentration ng/ml	Recovery (%)	ME (%)
Artificial tear solution	12.5	100.7	9.6	25	96.0	−10.3
	62.5	112.2	1.9	125	103.9	−12.2
Aqueous humor	12.5	89.8	4.9	25	97.2	10.2
	62.5	93.0	−3.8	125	85.5	6.9
Cornea	12.5	111.5	−5.3	25	103.9	5.9
	62.5	115.0	−9.1	125	105.6	9.1

Due to the high concentration of DEX/TOB in tear samples, a dilution integrity test was performed to extend the upper concentration limit of DEX and TOB by 500× dilution with the blank artificial tear solution. The standard solutions containing 62.5 μg/ml DEX and 125 μg/ml TOB were prepared and analyzed after 500 times dilution. The accuracy of DEX and TOB was 95.2% and 88.5%, respectively. The precision of DEX and TOB was 4.1% and 4.9%, respectively. Thus, the method can be used to test tear samples with DEX and TOB concentrations that exceed the calibration range.

Stability study results of DEX and TOB in aqueous humor, cornea and artificial tear solution using three replicates of QC samples at two concentrations under different storage conditions were presented in Table 4. The mean values of the analytes were found to be within ±15% deviation from the nominal concentration, indicating that no significant degradation of analytes was observed in matrices under the studied conditions, including the freeze-thaw stability at −20°C for 3 cycles with minimum of 24 h freezing between cycles and the autosampler stability at 10°C for 36 h.

Table 4.

Stability of DEX and TOB in artificial tear solution, aqueous humor and cornea under different conditions.

	Analyte	Nominal concentration ng/ml	DFN (%)
			Post-preparative (24 h in 4°C)	Freeze-thaw (three cycles at -20°C)
Artificial tear solution	DEX	12.5	−4.1	−6.9
		62.5	−8.5	−5.0
	TOB	25	−5.1	13.7
		125	9.3	1.1
Aqueous humor	DEX	12.5	1.4	8.2
		62.5	−7.8	−0.9
	TOB	25	−14.6	−6.4
		125	−5.9	6.2
Cornea	DEX	12.5	−1.4	8.5
		62.5	1.3	0.4
	TOB	25	−2.9	14.2
		125	−1.1	6.7

Due to the limited amount of blank aqueous humor and cornea samples, the analytical method of aqueous humor and cornea were partially validated. The QCs in aqueous humor and cornea were calculated under the calibration curves of artificial tear solution. The precision and accuracy were evaluated using aqueous humor/cornea at two concentrations with three replicates. The precision and accuracy of aqueous humor/cornea QCs are summarized in Table 5. The accuracy of aqueous humor and cornea were more than 90% of the nominal concentration, and the precision were within ±15%, which meets the acceptable limits of FDA guidance [12].

Table 5.

Partial validation of DEX and TOB in aqueous humor and cornea matrices with three replicates.

	Analyte	Nominal concentration ng/ml	Accuracy (DFN, %)	Precision (RSD, %)
Aqueous humor	DEX	12.5	102.0	12.0
		62.5	97.4	4.4
	TOB	25	90.1	5.2
		125	94.7	2.4
Cornea	DEX	12.5	96.6	11.1
		62.5	97.7	3.7
	TOB	25	90.9	5.1
		125	104.3	4.3

3.4. Pilot in vivo pharmacokinetic studies

We carried out a pilot ocular PK study over a period of 8 hours after topical administration of 50 mg Tobradex® ophthalmic ointment and successfully applied the developed method to quantify the DEX and TOB drug levels in rabbit tear fluids. The mean concentration versus time profiles of DEX and TOB are shown in Fig. 5C. T_max for both DEX and TOB was 15 mins after drug administration. DEX and TOB had a C_max of 32299.7 ng/ml and 18758.4 ng/ml, respectively, in rabbit tears after the topical ointment administration. The AUC_(0−∞) was 26626.9 ng/ml*h for DEX and 15586.6 ng/ml*h for TOB. In tear samples, DEX presented a higher conentration than TOB at the same time points, which demonstrated faster DEX release from the ointment to the tear solution.

Fig. 5:

Representative chromatograms of (A) DEX and DEX-d5, (B) TOB and sisomicin in rabbit tear fluids samples. (C) Concentration-time curve of DEX and TOB in rabbit tear fluids samples after topical administration of Tobradex® ointment. Mean ± SD, n=6 rabbit tear samples.

The ocular PK studies of Tobradex® ophthalmic ointment in human aqueous humor (NCT02734459) and Tobradex® ophthalmic suspension (0.3 wt% TOB and 0.1 wt% DEX) in rabbits have been previously conducted [5]. These studies mainly focused on the PK characterization of DEX without TOB. In the rabbit studies, DEX demonstrated 1.3±0.4 ng/ml in rabbit aqueous humor at 8 hours after topical administration of 30 μl of Tobradex® suspension containing 15 μg DEX [5]. In our pilot study, 50 mg Tobradex® ointment containing 50 μg DEX was administered. Because of the increased viscosity, opthalmic ointment can provide enhanced preocular retnetion and sustained drug release compared with ophthalmic suspensions that are usually quickly cleared from the ocular surface via blinking and tear turn-over [21]. The higher dose and longer retention from the Tobradex® ointment suggests a potentially higher DEX concentration in rabbit aqueous humor at 8 hours after topical Tobradex® ointment administration, indicating the ability of established method for the quantification of DEX in rabbit cornea and aqueous humor.

The limited corneal penetration of hydrophilic TOB and the requirment of highly sensitive quantification method might be the factors preventing the PK studies of TOB in ocular tissues, especially the aqueous humor and cornea. However, other non-corneal pathways that involves the drug transport via the conjunctiva and sclera and then into the intraocular tissues, can significantly contribute to the intraocular drug absoprtion for drugs with poor corneal permeability, such as gentamicin which has the similar structure as TOB [22]. The established method with LLOQ of TOB at 3 ng/ml could be helpful for the future PK studies of TOB in aqueous humor, cornea and other ocular tissues, which can facilitate the understanding of drug absorption mechanism of topical TOB ophthalmic products and the development of generic opthalmic formulations.

In this pilot study, we only used three rabbits (n=6 eyes) to continuously collet tear fluid samples at 8 different time points (15 min to 8 h) without the sacrifice of rabbits. Such tear fluid sample collection method may involve a potential loss of applied ophthalmic ointment during the animal study. Furthermore, we could not obtain the PK curve of drugs in rabbit cornea and aqueous humor through this pilot PK study. The main intent of this work is to establish an analytical method that is critical for our future detailed PK studies. The pilot ocular PK results of the tear solutions indicated the ability of the developed method to test the rabbit tear samples. And this method has been validated for DEX/TOB quantification in rabbit aqueous humor and cornea matrices which demonstrated the capability of the method to quantify DEX/TOB in rabbit cornea and aqueous humor. In our future studies, we will have n=3 rabbits (n=6 eyes) for each time point allowing only one-time sample collection to avoid the ointment loss because of repeated tear sample collection. At each time point, rabbits will be sacrificed to enucleate eyeballs immediately after tear fluid collection. Aqueous humor, cornea and other ocular tissues will be separated from the frozen eyeballs. The developed method will be used for quantification of DEX and TOB in rabbit aqueous humor, cornea and tear fluid samples and the intact PK paramters will be calculated.

4. Conclusion

A rapid, sensitive and selective LC-MS/MS bioanalytical method has been successfully developed for simultaneous quantification of DEX and TOB in rabbit ocular biofluids. A simple protein precipitation method was applied for DEX and TOB extraction from biological matrices. The LLOQ for DEX and TOB is 1.5 ng/ml and 3 ng/ml, respectively. The analytical method was successfully applied for a pilot ocular PK analysis of DEX and TOB in rabbit tears after topical administration of Tobradex® ointment to rabbit eyes. The developed method will be applied to quantify DEX and TOB drug levels in rabbit ocular fluids and tissues in the future studies for evaluating DEX/TOB ointment formulations.

Highlights.

• A bioanalytical method to simultaneously quantify dexamethasone and tobramycin.

• The sensitive, selective method was validated accordingly to FDA regulatory guidance.

• Application of the method was proven by an ocular PK study in rabbit tears.

• The method can be applied for ocular PK studies of dexamethasone and tobramycin.